Abstract
The concrete surface defects caused by various factors need to be repaired since the construction stage of the concrete dam. However, the binding strength between restored concrete and the original concrete will gradually get worse in the long run and appear some problems such as peeling. At present, the repaired concrete is mainly checked by the workforce, but it is time-consuming, inefficient, and hard to quantitative evaluate, such as the peeling area. A semantic segmentation method based on the DeepLabv3+ with ResNet50 backbone is proposed to restored concrete identification automatically. The dam restored concrete data set is established, including 372 high-resolution images to verify the method. The results indicated that the DeepLabv3+ model finally reaches 0.68 mIoU on the test set, which is a practical way to detect dam restored concrete.
Export citation and abstract BibTeX RIS
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.